In my previous blog on Saturday, I explored the growing reality that artificial intelligence is no longer confined to suggesting or supporting, but is increasingly able to act, initiating processes, shaping decisions, and raising fundamental questions of responsibility within care.
That discussion centred on accountability: on what it means when something acts in our name.
But there is another dimension now emerging.
Because even as we begin to address the question of action, something is unfolding beneath the surface, something that shapes those actions before they are ever taken.
This blog piece turns our attention to the unseen systems, influences and conditions that are already reshaping care.
Care practice is already being altered
We often assume that technological change in social care arrives through formal means: agreed systems, procurement processes, structured implementation. But practice has never evolved solely through formal routes.
Across workplaces, individuals are increasingly making use of Ai tools in informal ways, they are drafting reports, exploring ideas, clarifying complex information, and managing pressure. (Ellvero Insights, 2026; Aona AI, 2026). Such activity is often unrecorded, unshared, and unexamined. Yet it shapes how work is done.
What this represents is not simply adoption, but a subtle redistribution of influence within care practice itself.
And that matters. Because in a sector where accountability and transparency are essential, even small shifts in how decisions are reached begin to alter the fabric of professional responsibility.
At the same time, artificial intelligence is becoming part of the relational landscape of everyday life.
People are engaging with systems that respond immediately, adapt continuously, and offer a consistent form of interaction. Increasingly, these systems are used not only for information, but for reassurance, advice and emotional engagement (Andoh, 2026).
For social care, this raises deeper questions.
Care is rooted in relationship, that is human, imperfect, reciprocal. It involves presence, misunderstanding, negotiation, and hard graft.
If expectations of interaction begin to shift towards what is always responsive, always available and always consistent, then the context in which care is experienced may also begin to change. Not through replacement, but through comparison. And comparison, especially when unspoken, can reshape both expectation and experience.
Artificial intelligence often presents its outputs with clarity and confidence. But what sits beneath that clarity is dynamic.
These systems depend upon continually evolving datasets, external information flows, and complex training environments. As a result, outputs may shift over time, reflect inconsistencies, or evolve in response to changing inputs (Cyber.gov.au, 2026; Banker, 2025).
This means that what appears stable may in fact be fluid. For care, this introduces an important consideration. Because judgement relies not only on values and relationships, but on the interpretation of context.
If the knowledge environment itself is shifting, then maintaining confidence in judgement requires ongoing attentiveness.
Artificial intelligence is not simply software. It is sustained by an extensive and increasingly centralised infrastructure: data centres, global computing networks, complex supply chains and large-scale energy systems (World Economic Forum, 2025; S&P Global, 2025)
This infrastructure is shaped by forces that are economic, geopolitical and technical in nature.
Care, by contrast, remains local, anchored in communities, relationships, and place. This creates a tension and I feel that tension is likely to grow.
Because as care systems begin to depend upon Ai, they become connected to structures that are geographically distant, organisationally complex and governed beyond the immediate reach of care providers.
This raises significant questions of stewardship. How do local systems retain control of their direction? How do values and ethical standards translate across global infrastructures? Where does accountability sit when foundations are dispersed?
All this I would argue means that we urgently and with some focus need as a care system not just in Scotland but beyond need to rethink governance and assurance systems and processes.
All the developments I have noted above challenge traditional models of governance.
In the past governance and scrutiny has often focused on implementation, compliance and assurance. But emerging realities require something more. Because influence now occurs informally, invisibly, and dynamically.
Governance and scrutiny must therefore become more reflective, more adaptive and more attuned to practice as it evolves in real time. This may involve, for instance, creating organisational spaces for reflection, encouraging openness about emerging practice and supporting staff to question, not simply adopt. In this sense, governance and scrutiny shifts from being an exercise in control to a practice of shared awareness and it certainly involves much greater mutuality and collaboration than some of our traditional systems of scrutiny and governance have been enabled.
I think in addition that the changes I’ve noted around artificial intelligence also have profound implications for commissioning and leadership at a strategic level.
Artificial intelligence introduces new dependencies, for instance on infrastructure, on external providers, and on evolving systems. This requires a broader approach to commissioning.
Questions now extend beyond outcomes and cost to what dependencies are being created? From how resilient are the systems being relied upon to how adaptable are services over time?
Commissioning becomes not only transactional, but a form of long-term stewardship, attentive to interconnection, sustainability and risk (Davenport and Bean, 2026).
Lastly, for the workforce, these changes are immediate. Artificial intelligence shapes how work is experienced, how decisions are supported, and how professional identity is maintained. There is a need to ensure that staff understand the influences around their work, that they feel confident in their own judgement and critically that they are supported, not diminished, by technology
This is not only a question of training. It is a question of confidence, trust, and maintaining a sense of agency within evolving environments.
If these are not supported, there is a risk that change happens without inclusion, affecting the workforce rather than being shaped with it.
If the first phase of thinking about artificial intelligence in care has been about recognising its capacity to act, then this next phase asks us to recognise something more subtle. It asks us to attend to the systems that sit beneath action, the influences that shape judgement and the environments that are quietly forming around us
Not all change is visible. Not all transformation is immediate. But in social care, where attentiveness to the human condition is central, what we do not see can matter as much as what we do. Because the future of care will not be defined simply by what artificial intelligence enables. It will be defined by how we understand the conditions it creates and whether, within those conditions, we remain fully present to what it means to care.
Donald Macaskill
References.
Aona AI (2026) Shadow AI statistics 2026: risks and governance. Available at: https://aona.ai/blog/shadow-ai-statistics-2026/
Andoh, E. (2026) ‘AI chatbots and digital companions are reshaping emotional connection’, Monitor on Psychology, 57(1). Available at: https://www.apa.org/monitor/2026/01-02/trends-digital-ai-relationships-emotional-connection
Banker, S. (2025) ‘AI risks include data poisoning and model corruption’, Forbes. Available at: https://www.forbes.com/sites/stevebanker/2025/01/15/ai-risks-include-data-poisoning-and-model-corruption/
Cyber.gov.au (2026) Artificial intelligence and machine learning: supply chain risks and mitigations. Available at: https://www.cyber.gov.au
Davenport, T.H. and Bean, R. (2026) ‘Five trends in AI and data science for 2026’, MIT Sloan Management Review. Available at: https://sloanreview.mit.edu/article/five-trends-in-ai-and-data-science-for-2026/
Ellvero Insights (2026) Shadow AI in the enterprise: the hidden risk most leaders underestimate. Available at: https://www.ellvero.com
S&P Global (2025) AI’s global resource race: challenges and opportunities. Available at: https://www.spglobal.com
World Economic Forum (2025) AI geopolitics and data centres in the age of technological rivalry. Available at: https://www.weforum.org
